A place-focused model for social networks in cities

TL;DR

This paper introduces a model for city-based social networks based on the focused organization theory, utilizing large-scale location data to analyze how meeting places influence friendship formation and network structure.

Contribution

It presents a novel model for city social networks that incorporates place categories, supported by empirical analysis of large-scale location-based social data.

Findings

City social networks share common structural properties.

Place categories significantly influence friendship likelihood.

The proposed model replicates observed network structures.

Abstract

The focused organization theory of social ties proposes that the structure of human social networks can be arranged around extra-network foci, which can include shared physical spaces such as homes, workplaces, restaurants, and so on. Until now, this has been difficult to investigate on a large scale, but the huge volume of data available from online location-based social services now makes it possible to examine the friendships and mobility of many thousands of people, and to investigate the relationship between meetings at places and the structure of the social network. In this paper, we analyze a large dataset from Foursquare, the most popular online location-based social network. We examine the properties of city-based social networks, finding that they have common structural properties, and that the category of place where two people meet has very strong influence on the likelihood of their being friends. Inspired by these observations in combination with the focused organization theory, we then present a model to generate city-level social networks, and show that it produces networks with the structural properties seen in empirical data.